Methane point-source attribution from orbit
Satellite spectrometers now detect discrete methane plumes from oil and gas infrastructure, landfills, and coal mines. Resolution and detection floors vary sharply between sensors, and choosing the wrong one produces false assurance.
Sensors
- Sentinel-5P TROPOMI: Hyperspectral push-broom spectrometer covering SWIR near 2305–2385 nm. Nadir pixel footprint approximately 5.5 × 7 km since August 2019 (improved from original 7 × 3.5 km). Daily global coverage. Detection floor for a single overpass is roughly 500 kg/hour for an isolated point source; useful as a basin-scale screening tool, not a per-facility compliance instrument.
- GHGSat-D/C series: Commercial Fabry-Pérot imaging spectrometers targeting the SWIR methane absorption band near 1630 nm. Spatial resolution approximately 25 m per pixel, enabling individual well-pad attribution. Minimum detectable emission rate published by GHGSat at roughly 10–25 kg/hour under favourable wind and albedo conditions. Tasked on demand; not a continuous global monitor.
- EMIT (ISS-mounted imaging spectrometer, Carbon Mapper heritage): NASA EMIT on the International Space Station covers 380–2500 nm at approximately 60 m spatial resolution. Designed for surface mineralogy but demonstrated methane plume detection at large point sources. Opportunistic coverage tied to ISS orbit; not taskable on demand. Carbon Mapper's dedicated constellation will build on this heritage.
- Sentinel-2 MSI (proxy retrieval): Multispectral imager with a band centred near 2190 nm (Band 12, 20 m resolution). Not a dedicated methane sensor, but differential absorption between Band 11 and Band 12 allows proxy retrievals for very large plumes. Detection floor is high, roughly several thousand kg/hour, and results carry significant albedo-related uncertainty. Revisit is 5 days at the equator.
Why methane is hard to see and easy to miss
Methane absorbs sunlight in two shortwave-infrared windows: near 1630 nm and near 2300 nm. A spectrometer looking down through the atmosphere measures the total column of methane between the satellite and the surface. The signal from a single leaking compressor is tiny against that background column. Isolating it requires either very high spatial resolution, so the plume fills a meaningful fraction of the pixel, or very precise spectral calibration that can detect column enhancements of a few parts per billion against a background of roughly 1900 ppb.
Wind complicates everything. A plume disperses rapidly, and emission-rate estimates derived from satellite data depend on wind speed at the moment of overpass. Published studies using TROPOMI data typically apply reanalysis winds from ERA5 or MERRA-2, which carry their own uncertainty, often 10–30 percent. At GHGSat resolution, analysts can sometimes measure the plume's spatial extent directly and use the integrated mass enhancement method to reduce wind-model dependence, but the approach still requires a wind estimate.
What TROPOMI can and cannot tell a regulator
TROPOMI's daily global pass is genuinely valuable at the basin scale. Studies published in journals such as Science and Nature have used TROPOMI to identify entire oil-producing regions, including the Permian Basin in Texas and the Hassi Messaoud field in Algeria, as emitting far more methane than national inventories suggested. That kind of systematic, country-level discrepancy is exactly what TROPOMI was built to expose.
The limit is spatial. At 5.5 × 7 km per pixel, a single overpass cannot reliably attribute a plume to one facility among several clustered together. The published detection floor of roughly 500 kg/hour means small but chronic leaks, a poorly sealed valve, a malfunctioning pneumatic controller, go undetected. Persistent aggregation of TROPOMI overpasses over weeks can push the effective detection limit lower, but at the cost of temporal resolution. TROPOMI is a screening tool. It tells you where to look; it does not tell you which pipe to fix.
What a 25-metre pixel changes
GHGSat's commercial imagers work at approximately 25 m resolution, which is fine enough to isolate individual infrastructure components: a specific tank battery, a compressor station, a coal-mine ventilation shaft. The integrated mass enhancement (IME) method sums the methane column enhancement across all pixels within the detected plume, multiplies by an effective wind speed, and divides by the plume's effective length to estimate a flux. Published GHGSat analyses have quantified sources as small as 10–25 kg/hour under good observing conditions.
The catch is coverage. GHGSat tasks individual scenes on request; it does not provide continuous global monitoring. A facility that leaks intermittently may not be emitting during the tasked overpass. Combining TROPOMI screening, which flags anomalous basins every day, with targeted GHGSat tasking over the flagged area is the approach that makes operational sense. Neither sensor alone closes the attribution loop.
The cloud problem, and what honest revisit figures mean
SWIR retrievals require reflected sunlight from the surface. Cloud cover blocks the signal entirely. In persistently cloudy regions, such as the Niger Delta or parts of Indonesia, useful TROPOMI retrievals may be available only on 30–50 percent of days. GHGSat faces the same physics. This is not a solvable problem with current passive sensors; it is a hard physical constraint that any monitoring programme must account for when designing compliance schedules.
Revisit figures quoted in sensor specifications describe how often the satellite passes over a location, not how often a usable retrieval is obtained. Effective revisit, accounting for cloud, is always lower and varies by season and geography. Polar-orbiting sensors such as TROPOMI achieve one overpass per day at most latitudes, but a given facility in a cloudy region might yield only a handful of cloud-free retrievals per month.
Turning a plume into a number: the flux estimation chain
The integrated mass enhancement method is the most widely used approach for point-source quantification from high-resolution imagery. The analyst identifies the plume boundary, typically using a threshold enhancement above background. They sum the column enhancement in each pixel, convert to a mass of methane per unit area, and integrate spatially to get a total mass in the plume at the moment of observation. Dividing by the plume's downwind length and multiplying by wind speed gives a flux in kilograms per hour.
Uncertainty in that final number is substantial. Published analyses typically report one-sigma uncertainties of 30–60 percent, driven mainly by wind uncertainty, plume boundary ambiguity, and surface albedo variation. That is honest science, but it creates friction in enforcement contexts where regulators want defensible, legally admissible measurements. Satellite data currently sits most comfortably in the roles of screening, prioritisation, and trend monitoring, rather than as a primary measurement for penalty proceedings. Ground-based or airborne verification remains the standard for enforcement at the individual-facility level.
Satellize runs TROPOMI column-retrieval workflows and can layer in commercial tasking through client licence arrangements. The Tonga crop-estimation programme demonstrated the organisation's approach to uncertainty-bounded quantitative outputs; the same statistical discipline applies here.
Designing a monitoring programme that will survive scrutiny
A credible methane monitoring programme needs to specify, in advance, what detection threshold it accepts, how it handles cloud-contaminated overpasses, and what wind-data source it uses for flux estimation. These choices should be documented before data collection begins, not chosen post hoc to produce convenient numbers.
The practical architecture for most national or basin-level programmes starts with daily TROPOMI screening to identify persistent anomalies. Facilities or clusters that exceed a defined column-enhancement threshold trigger tasked high-resolution observations from GHGSat or equivalent. Quantified plumes above a second threshold trigger ground or airborne follow-up. That three-tier structure is consistent with how the European Union's LULUCF and methane regulations are beginning to frame satellite-based verification, and it reflects the honest capabilities and limits of current sensors.
Typical figures
| TROPOMI spatial resolution | 5.5 × 7 km per pixel (since August 2019) |
| GHGSat spatial resolution | ~25 m per pixel |
| EMIT / Carbon Mapper resolution | ~60 m per pixel |
| TROPOMI revisit | Daily global coverage (one overpass per day at most latitudes) |
| GHGSat revisit | On-demand tasking; not a continuous monitor |
| TROPOMI minimum detectable emission rate | ~500 kg/hour for a single overpass at an isolated point source |
| GHGSat minimum detectable emission rate | ~10–25 kg/hour under favourable conditions (published by GHGSat) |
| Key spectral bands | SWIR ~1630 nm (GHGSat, EMIT); SWIR ~2305–2385 nm (TROPOMI) |
| Flux estimation uncertainty | Typically 30–60% one-sigma, dominated by wind and plume-boundary uncertainty |
| TROPOMI archive depth | From May 2018 (launch); data freely available via Copernicus Dataspace |
Analytics Satellize can run
| Basin-scale methane anomaly screening | TROPOMI XCH4 column retrieval; multi-day aggregation to reduce noise floor | Monthly ranked list of anomalous grid cells with column-enhancement values, delivered as GIS layer and PDF summary |
| Point-source plume detection and attribution | High-resolution SWIR imagery (GHGSat or equivalent); plume segmentation against background column | Per-facility plume detection report with georeferenced plume boundary and attributed infrastructure ID |
| Emission rate quantification | Integrated mass enhancement (IME) with ERA5 wind input; uncertainty propagation reported | Flux estimate in kg/hour with stated one-sigma uncertainty range, per detected event |
| Persistent emitter ranking | Time-series aggregation of TROPOMI overpasses; frequency-weighted emission scoring | Quarterly persistent-emitter register for a defined basin or country, in spreadsheet and GIS format |
| Cloud-gap analysis and effective revisit reporting | Quality-flag filtering of TROPOMI L2 product; cloud-fraction threshold applied per overpass | Per-facility table of usable observation days per month, enabling realistic monitoring-programme design |
| Sentinel-2 proxy retrieval for very large sources | Band 11 / Band 12 differential absorption retrieval; uncertainty bounds flagged explicitly | Alert layer for column enhancements above ~3000 kg/hour threshold, with explicit caveat on albedo sensitivity |
Who does the work
We can get this done for you. Satellize runs its own analyst desk and a strong science team. You do not buy a data feed and work out what it means; our people source the imagery, run the analysis described on this page, and hand you the answer with its confidence limits stated. Discuss this requirement.